The sawtooth oscillator should be at full amplitude the first half cycle
of the tone. And the start up chirp done in that time period. It won't
hang over either when unkeyed.
The twin-T undoubtedly takes up to a couple dozen cycles to start and
changes frequency during that.
Any filter causes some delay, the more samples the more delay in a DSP
filter, but the better the relative time response. E.g. the less
difference of time delay according to frequency.
I sometimes call CW with my DSP-599+ in the speaker line. I've not
looked for delay, but something seems not quit right though the tone
quality is super, even on SSB mode. Maybe next time I rattle off a call
at 25 wpm, I'll see if there's some time delay to bug me. I know there's
enough time delay that I'd rather tune the band with the DSP bypassed.
A more optimum filter would be a passive lowpass filter in the speaker
line. It could have a cutoff of 2.5 KHz or so, and would clean up the
side tone with less delay than a DSP and would contribute NO noise to
the signal while cleaning up hiss from the radio. I have created a
couple that have switched cutoffs and published them locally. They work.
When on CW using a tone of about 400 Hz and cranking the LP filter down
to about 450 Hz cutoff, makes signals off the air sound like a super
high quality local code practice oscillator. No harmonics, and very
little noise. I should maybe post that design to my web page someday
along with my power supply design and other good ideas.
73, Jerry, K0CQ
--
Entire content copyright Dr. Gerald N. Johnson. Reproduction by
permission only.
|